Skip to content

[SPARK-20098][PYSPARK] dataType's typeName fix #17435

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 9 commits into from
Closed

[SPARK-20098][PYSPARK] dataType's typeName fix #17435

Changes from 2 commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
00994e9
getType fix
szalai1 Mar 26, 2017
933f3cb
type
szalai1 Mar 26, 2017
0a18192
StructField's typeName raise TypeError
szalai1 May 13, 2017
6bf11f0
add test
szalai1 May 20, 2017
8872e19
fixing pep8 style issues
szalai1 Jun 25, 2017
a18436d
test failure fixed
szalai1 Sep 7, 2017
a451901
improving error message
szalai1 Sep 8, 2017
98bac12
adding space
szalai1 Sep 8, 2017
28f142c
space at thr right place
szalai1 Sep 9, 2017
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 19 additions & 1 deletion python/pyspark/sql/types.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,7 +57,25 @@ def __ne__(self, other):

@classmethod
def typeName(cls):
return cls.__name__[:-4].lower()
typeTypeNameMap = {"DataType": "data",
"NullType": "null",
"StringType": "string",
"BinaryType": "binary",
"BooleanType": "boolean",
"DateType": "date",
"TimestampType": "timestamp",
"DecimalType": "decimal",
"DoubleType": "double",
"FloatType": "float",
"ByteType": "byte",
"IntegerType": "integer",
"LongType": "long",
"ShortType": "short",
"ArrayType": "array",
"MapType": "map",
"StructField": "struct",
Copy link
Member

@HyukjinKwon HyukjinKwon Mar 27, 2017 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It seems this problem only applies to StructField. Could we just override typeName with simply throwing an exception? I think users are not supposed to call typeName against StructField but simpleString against the instance.

BTW, It apparently seems a bit odd that it extends DataType though.. I guess probably some tests are broken if we change the parent as it seems it is dependent on the parent assuming from #1598. So, I guess minimised fix would be just to override it.

Copy link
Member

@HyukjinKwon HyukjinKwon Mar 27, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

cc @holdenk and @viirya WDYT?

Copy link
Member

@viirya viirya Mar 27, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah, I don't think it is valid to call typeName against a StructField. Actually, StructField is not a data type, strictly speaking...

I don't know why StructField inherits DataType in pyspark. In scala, it is not.

Throwing an exception when calling typeName on StructField seems good enough, instead of a map like this.

Copy link
Contributor Author

@szalai1 szalai1 Mar 27, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@viirya The way I found this bug:
I wanted to figure out the schema of a dataset. I loaded it into a data frame and asked its schema. Then I called the typeName on each column. I do not know this is/was best way to do this, but I think it is valid to call typeName against a dataType to get its real type.

Copy link
Member

@HyukjinKwon HyukjinKwon Mar 27, 2017 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It is valid call for DataTypes but I don't think it is against StructField. If you look at the Scala-side codes, StructField is not a DataType (which is "The base type of all Spark SQL data types") but it seems the parent became DataType in Python for some reason in the PR I pointed out. In any way, It seems "A field inside a StructType".

If you want to know the schema, you could simply

>>> spark.range(1).schema[0].simpleString()
'id:bigint'
>>> spark.range(1).schema.simpleString()
'struct<id:bigint>'
>>> type(spark.range(1).schema[0])
<class 'pyspark.sql.types.StructField'>
>>> str(spark.range(1).schema[0])
'StructField(id,LongType,false)'

Could you elaborate your use-case and why simpleString is not enough?

Copy link
Member

@viirya viirya Mar 29, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@szalai1 I think @HyukjinKwon 's code snippets should address your request. Doesn't it?

Copy link
Member

@viirya viirya Mar 29, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Btw, I don't think i.typeName() is a valid usage. We better let it throw an exception when calling typeName on StructField.

i.dataType.typeName() is more reasonable call to me.

Copy link
Contributor Author

@szalai1 szalai1 Mar 29, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@viirya You are right. I will use it that way. Thanks @HyukjinKwon !

But my bug report is still valid, I think. Can I override the typeName function?

Copy link
Member

@HyukjinKwon HyukjinKwon Mar 29, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yup, I think we should still fix and overriding it is good enough.

Copy link
Member

@viirya viirya Mar 29, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

+1

"StructType": "struct"}
return typeTypeNameMap[cls.__name__]

def simpleString(self):
return self.typeName()
Expand Down